Mathematical models of safety monitoring system considering uncertainty

Abstract

Statistical decision models have been used successfully in the past to consider the design problem of safety monitoring system. The design problem aims to decide a global optimization of a distributed sensor detection system. The Baysian models are effective but not adequate for analyzing uncertainty in the limits of sensors so that two models are proposed. 1. The first probabilistic model aims to reduce the uncertainty by a combination of two types of sensors that have different thresholds, the SP sensor and the FW sensor. 2. The other is a model based on Dempster-Shafer theory that includes probability theory as a special case. This model easily expresses uncertain information, and in the model the optimal combination rule for coordinating contradicted information is proposed. The latter model is a general state-action model so that it is expected to utilize for the mathematical analysis of human-machine system.